Three dimensional (3D) video applications are becoming increasingly common as they may provide a more compelling visual experience to the viewer. One method of generating 3D video makes use of multiple cameras to derive depth information based on the disparity resulting from a perspective offset between the cameras, in a manner similar to human binocular vision. Conventional techniques typically estimate this disparity based on an analysis of corresponding still image video frames from each camera. This per-frame approach, however, often results in unpleasant artifacts such as flickering between consecutive frames. Some techniques have been developed to improve disparity estimation but they generally impose large computational burdens or excessive memory requirements, or they rely on limiting assumptions about the video characteristics.
Although the following Detailed Description will proceed with reference being made to illustrative embodiments, many alternatives, modifications, and variations thereof will be apparent to those skilled in the art.